ASN's Mission

To create a world without kidney diseases, the ASN Alliance for Kidney Health elevates care by educating and informing, driving breakthroughs and innovation, and advocating for policies that create transformative changes in kidney medicine throughout the world.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on X

Kidney Week

Please note that you are viewing an archived section from 2023 and some content may be unavailable. To unlock all content for 2023, please visit the archives.

Abstract: TH-PO016

Computed Tomography Radiomics Analysis for Discrimination and Severity Assessment of Diabetic Kidney: A Retrospective Machine Learning Study

Session Information

Category: Augmented Intelligence, Digital Health, and Data Science

  • 300 Augmented Intelligence, Digital Health, and Data Science

Authors

  • Chu, Seung Hye, Soonchunhyang University Hospital, Seoul, Korea (the Republic of)
  • Kim, Soyeon, Soonchunhyang University Hospital, Seoul, Korea (the Republic of)
  • Noh, Hyunjin, Soonchunhyang University Hospital, Seoul, Korea (the Republic of)
  • Kwon, Soon hyo, Soonchunhyang University Hospital, Seoul, Korea (the Republic of)
Background

Kidney radiomics has been evaluated for the development of accurate diagnostics tools for renal tumors. However, there is a scarcity of radiomics studies focused on diabetic kidney disease (DKD). In this study, we aimed to investigate whether computed tomography (CT) radiomics features can differentiate DKD from normal kidneys and assess the severity of DKD.

Methods

We analyzed type 2 diabetes mellitus (T2DM) patients and healthy controls (HCs) who underwent abdominal CT scans between November 2014 and November 2022. CT volumetric data of both kidneys were extracted using a deep-learning model, enabling radiomics feature extraction. T2DM patients were categorized into risk groups based on estimated glomerular filtration rate (eGFR) and degree of albuminuria. Machine learning (ML) models were used to differentiate DKD patients from HCs and classify DKD risk groups. The models were trained and evaluated on separate patient data sets, with performance metrics such as sensitivity, specificity, accuracy, and area under the curve (AUC).

Results

The study included 462 T2DM patients and 90 HCs, who were randomly assigned to a training set (n=386; mean age ± standard deviation, 60.9 years ± 16.2; 239 men) or a test set (n=166; mean age, 60.7 years ± 15.7; 91 men). A total of 1,219 radiomics features were extracted. The random forest model showed excellent performance in differentiating between HCs and patients with low-risk DKD, with an AUC of 1.00 and an accuracy of 100% in the training set and an AUC of 0.84 and an accuracy of 80.6% in the test set. It also showed a good performance in discriminating between DKD groups based on eGFR (AUCs, 0.99-1.00, in the training set; AUCs, 0.70-0.94, in the test set) (Table1).

Conclusion

CT-derived radiomics analysis of the kidneys can effectively differentiate diabetic kidneys from normal kidneys and assess the severity of DKD. These findings suggest that radiomics data capture pathological changes in the kidneys associated with diabetes.

Funding

  • Government Support – Non-U.S.